摘要
将认知无线电系统中的传输调度方案建模为一个约束马尔科夫决策过程(CMDP),即在满足缓存器内包数约束的情况下最小化发送数据包消耗的平均功率。因为在认知无线电系统中,环境参数无法预先得知,为此利用R学习来自适应地获取CMDP的近似最优策略。在仿真结果中,对基于R学习的传输调度方案的性能进行了比较和分析,结果显示该方案能适用于参数未知的环境且有效地降低平均功率。
Transmission and scheduling scheme of average power minimization under the constraint of the number of packets in buffer is addressed as a constrained Markov decision process （CMDP）. The environment parameters in cognitive systems could not known in advance, soRlearningis utilized to adaptively achieve thenearlyoptimal policy. Simulation results are given, thus to evaluate the performances of R learning-based scheme. These results show that the scheme adapts to the parameters-unknown environment and could reduce the average power effectively.
出处
《通信技术》
2009年第10期23-25,28,共4页
Communications Technology
基金
重庆邮电大学自然科学基金资助项目(编号A2007-18)
关键词
认知无线电
R学习
传输调度
cognitive radio
R learning
transmission and scheduling
